diff --git a/Misc/NEWS.d/next/Core_and_Builtins/2025-06-24-16-46-34.gh-issue-135904.78xfon.rst b/Misc/NEWS.d/next/Core_and_Builtins/2025-06-24-16-46-34.gh-issue-135904.78xfon.rst new file mode 100644 index 00000000000000..ecbd8fda9a5e9d --- /dev/null +++ b/Misc/NEWS.d/next/Core_and_Builtins/2025-06-24-16-46-34.gh-issue-135904.78xfon.rst @@ -0,0 +1,2 @@ +Perform more aggressive control-flow optimizations on the machine code +templates emitted by the experimental JIT compiler. diff --git a/Tools/jit/_optimizers.py b/Tools/jit/_optimizers.py new file mode 100644 index 00000000000000..1077e4106fdfbd --- /dev/null +++ b/Tools/jit/_optimizers.py @@ -0,0 +1,319 @@ +"""Low-level optimization of textual assembly.""" + +import dataclasses +import pathlib +import re +import typing + +# Same as saying "not string.startswith('')": +_RE_NEVER_MATCH = re.compile(r"(?!)") +# Dictionary mapping branch instructions to their inverted branch instructions. +# If a branch cannot be inverted, the value is None: +_X86_BRANCHES = { + # https://www.felixcloutier.com/x86/jcc + "ja": "jna", + "jae": "jnae", + "jb": "jnb", + "jbe": "jnbe", + "jc": "jnc", + "jcxz": None, + "je": "jne", + "jecxz": None, + "jg": "jng", + "jge": "jnge", + "jl": "jnl", + "jle": "jnle", + "jo": "jno", + "jp": "jnp", + "jpe": "jpo", + "jrcxz": None, + "js": "jns", + "jz": "jnz", + # https://www.felixcloutier.com/x86/loop:loopcc + "loop": None, + "loope": None, + "loopne": None, + "loopnz": None, + "loopz": None, +} +# Update with all of the inverted branches, too: +_X86_BRANCHES |= {v: k for k, v in _X86_BRANCHES.items() if v} + + +@dataclasses.dataclass +class _Block: + label: str | None = None + # Non-instruction lines like labels, directives, and comments: + noninstructions: list[str] = dataclasses.field(default_factory=list) + # Instruction lines: + instructions: list[str] = dataclasses.field(default_factory=list) + # If this block ends in a jump, where to? + target: typing.Self | None = None + # The next block in the linked list: + link: typing.Self | None = None + # Whether control flow can fall through to the linked block above: + fallthrough: bool = True + # Whether this block can eventually reach the next uop (_JIT_CONTINUE): + hot: bool = False + + def resolve(self) -> typing.Self: + """Find the first non-empty block reachable from this one.""" + block = self + while block.link and not block.instructions: + block = block.link + return block + + +@dataclasses.dataclass +class Optimizer: + """Several passes of analysis and optimization for textual assembly.""" + + path: pathlib.Path + _: dataclasses.KW_ONLY + # prefix used to mangle symbols on some platforms: + prefix: str = "" + # The first block in the linked list: + _root: _Block = dataclasses.field(init=False, default_factory=_Block) + _labels: dict[str, _Block] = dataclasses.field(init=False, default_factory=dict) + # No groups: + _re_noninstructions: typing.ClassVar[re.Pattern[str]] = re.compile( + r"\s*(?:\.|#|//|$)" + ) + # One group (label): + _re_label: typing.ClassVar[re.Pattern[str]] = re.compile( + r'\s*(?P